Sesame releases a demo of an open source AI voice model that can converse naturally with humans with amazing performance and quality, and anyone can try it for free
AI startup Sesame has developed a voice AI model that speaks as naturally as a human, and has released a demo that anyone can use, so I tried it out.
Crossing the uncanny valley of conversational voice
https://www.sesame.com/research/crossing_the_uncanny_valley_of_voice
This AI's human-like voice raises emotional attachment fears
https://techissuestoday.com/this-ais-human-like-voice-raises-emotional-attachment-fears/
At Sesame, we believe in a future where computers are lifelike. Today we are unveiling an early glimpse of our expressive voice technology, highlighting our focus on lifelike interactions and our vision for all-day wearable voice companions. https://t.co/Edp8V8urgC pic.twitter.com/Mc5nWnBJZM
— Sesame (@sesame) February 27, 2025
When you open the demo page , it looks like this. It seems that you can talk to two AIs, 'Maya' and 'Miles.' Click 'Maya' this time.
You will be asked for permission to use the microphone, so click 'Allow this time only.'
The conversation started. Maya suddenly started speaking to me. Just like Sesame sings, she spoke like a human, and with the level of English I learned at school, she spoke so naturally that I panicked and wondered if I'd accidentally spoken to a native speaker.
The actual conversation goes like this:
After the conversation is over, you can download the conversation by clicking 'Download clip.'
On Hacker News, a news site for engineers, people who use a lot of English on a daily basis
commented
on Sesame's AI.'I tried the demo and was really surprised at how human-like it was. I'm very excited to be able to use a voice assistant like this, and a little worried that I might start to empathize with a voice assistant that sounds this human-like,' said one commenter . However, opinions were divided, with one British commenter, noodlesUK, saying that he was 'very uncomfortable with the level of artificial cheerfulness in the voice' and that 'the whole experience is strange and unsettling .'
Sesame's Brendan Iribe also appeared on Hacker News , writing, 'Despite the challenges, we're confident we can achieve a natural, conversational interface.'
Sesame plans to open source this model and release it on GitHub . At the time of writing, it was stated that work to make it public was underway.
Related Posts:
